Data & Analytics Solutions
Unifying 6 Data Sources into a Single Business Intelligence Layer
The retailer was flying blind — sales, inventory, logistics, and marketing data lived in six different systems with no shared reporting layer.
- Client
- A multi-channel retail group
- Industry
- Retail & E-Commerce
- Duration
- 10 weeks
- Year
- 2024
The challenge
Leadership was making pricing and stocking decisions based on week-old reports extracted manually from SAP, Shopify, Google Ads, and three regional logistics providers. With 200+ SKUs and seasonal spikes, the lag was causing both overstock and stockout events simultaneously.
Our approach
We designed an ETL pipeline using Python and Apache Airflow that pulls data nightly from all six sources, normalises it into a star schema in Redshift, and feeds a Power BI workspace. We also built custom anomaly detection scripts that fire Slack alerts when any KPI deviates more than 15% from its 30-day rolling average.
Results
- Reporting turnaround dropped from 5 days to same-day
- Overstock incidents reduced by 34% in the first quarter post-launch
- CEO dashboard refresh reduced from weekly manual effort to a 30-second automated run
- Anomaly alerts caught a logistics pricing error worth €28K within the first month
Technology
- Python
- Apache Airflow
- AWS Redshift
- Power BI
- PostgreSQL
- Slack API
This project was delivered as part of our data & analytics solutions service.